> It seems like we're being inconsistent here in allowing 'where xid = > integer' but not allowing 'where xid != integer'.
Well, if you look into pg_operator you'll soon find that there are exactly two built-in operators that accept type xid: "=(xid,xid)" and "=(xid,integer)" (where I'd say the latter is just a kluge). There hasn't previously been any demand to flesh it out more than that. Do you have an actual use-case where <> would be helpful, or is this just experimentation?
I'm not sure yet. I was doing some thinking about ways to do incremental backups (at least for inserted/updated rows, deleted rows present a different challenge), and was just doing some simple queries to see what worked and what didn't..
It also appears you cannot group on a column of type xid.
Would adding a <> operator enable that? -- Mike Nolan